Surface defect detection of OLED display is difficult due to the characteristics of periodic texture background, fuzzy defect boundary and low contrast. In order to solve this problem, some investigations are carried out for automatic surface defect detection for the OLED display in this paper. The singular value decomposition (SVD) method is implemented on the OLED image, and the first two larger singular values are selected to reconstruct the image texture background. The differential operation between the original image and the reconstructed image is carried out to obtain the residual image. The initial membership value is randomly set to every pixel of the residual image, and the final membership value is obtained by the fuzzy c-means (FCM) clustering method. Based on the final membership value of every pixel, the pixels are grouped into two categories to segment the defects accurately from the residual image. The periodic texture background of OLED display screen can be reconstructed effectively by selecting two larger singular values, and the average U value obtained by the FCM method is 0.9846. The method of background reconstruction based on SVD can effectively detect the surface defects of OLED display. Compared with the watershed method and the Otsu method, the FCM method can accurately segment the defect areas of fuzzy boundary.